Student Capstone Project

Team building and technical know-how.

Students in the M.Eng. in Engineering program will demonstrate their proficiency through a team-based design project. Project ideas are proposed by clients from industry, teaching hospitals, and clinicians seeking solutions to specific problems. Student teams assess the market and conduct competitive analysis, engineering design, software development, proto-typing, testing and documentation of results.  Weekly or biweekly update meetings with clients are essential to the success of the project.  Teams are expected to self-organize their effort by assigning tasks, developing a schedule, identifying bottlenecks, and gathering resources.

Working with the clients, the teams are expected to gain insights to help them implement their idea. During the project, the teams may request guidance from program faculty and may take field trips to the client’s location. Project presentations and demonstrations are delivered during a formal end-of-program event.

For companies looking to engage our M.Eng. students on a capstone project, please submit a project intake form

Here are some project examples:

2019 Capstone Projects

(Client: Professor, Health Care Systems Engineering)
A portable ultrasound imaging-based breast biopsy system

Advances in diagnostic devices for biopsies is limited to better needles and separately to tissue capturing systems. There have been few developments in integrated systems combining imaging and the biopsy procedure on a single platform. In recent years, MRI-based integrated systems have seen some innovation, but there is still a need for a modern stream-lined system that can accurately identify and localize target regions for breast biopsy.

 

(Client: Medical device company)
Microscopy instrumentation for nerve identification

Transdermal and intraoperative identification and differentiation
of nerves from vasculature. The project also involved the review
and critique of the current state-of-the-art of light technology
for human nerve visualization.

 

(Client: Global medical device company)
Machine learning based physiological signal monitoring during clinical imaging scans
Physiological data from a patient is a vital tool for medical diagnostics since it holds invaluable information reflecting the patient’s health status. Monitoring physiological signals could assist in the decision-making, and selection of scan modality and protocol parameters in the clinical setting.

This project aimed to develop a smart tool that automatically optimizes imaging strategies using deep learning.

 

(Client:  Regional medical hospital)
Intra-abdominal biodegradable amylase sensor

Postoperative pancreatic fistula (PODF) is the most common and dangerous complication of pancreatic surgery, affecting 13% to 41% of patients. Surgically placed drains to detect pancreatic fistula often cause intra-abdominal infection and pain in the abdomen.  Early detection and management of pancreatic fistula are very important.

The objective of this capstone project was to develop a biodegradable and implantable sensor.

 

(Client:  Biopharmaceutical company)
Improved medication delivery device or process
This project involved the review and analysis of the current medication delivery methods such as IV infusion, push, pumps, intramuscular, etc.  The team assessed the considerations, requirements, decision making, and processes involved with medication.  They also identified innovative concepts to address unmet needs and prototyped a bed-side medication delivery pump.

 

(Client: Global medical device company)
Deep learning for brain anatomy segmentation

With the rapid development of the medical instrumentation field, MRI plays more important roles in the diagnosis of brain diseases. The study of the different structures of brain is essential in the diagnosis and treatment of diseases such as Alzheimer Disease (AD) and Parkinson’s Disease (PD). The two main brain segmentation methods currently used (manual segmentation and software segmentation) are time-consuming, inefficient and complicated.

The objective was to develop deep learning architecture and its optimization for medical image segmentation and classification.

 

(Client: Global medical device company)
Multiscale contrast enhancement for MRI imaging
Magnetic resonance images usually contain both large contrast variations and small vital low contrast details. Applying postprocessing could be helpful to satisfy the conflicting needs of reproducing the low contrast details while maintaining the general gray value range.  A multiscale method, especially an image pyramid, has proven to be a very versatile and efficient algorithm when applied to other kinds of images.  This project objective was to explored the application of the LPSVD (Laplacian pyramid combined with SVD) algorithm to enhance MR images.  This could lead to greater image amplification of the vital areas, while minimizing background “noise”.

 

(Client: Global medical device company)
Motion artifact through head motion tracking during MRI

High-resolution magnetic resonance imaging (MRI) requires prolonged scan time to maximize spatial resolution, therefore, this imaging modality is highly sensitive to artifacts caused by motion during the scanning process.  Subject physiologic motions such as blood flow, respiratory and cardiac motion, and gross movements can create undesirable phase shifts that commonly result in image blur or the presence of “ghosts”.

The objective of the project was to develop a motion tracking toolkit capable of tracking head movement within a position matrix.

 

(Client: Biopharmaceutical company)
Improved patient critical care
The project involved identifying opportunities to improve the diagnostic and therapeutic procedures of critical care patients
with Acute Kidney Injury (AKI) being treated with continuous renal replacement therapy (CRRT).

 

(Client:  Biopharmaceutical company)
Progressive Supranuclear Palsy screening battery

Progressive Supranuclear Palsy (PSP) is a rare, progressive, ultimately fatal neurological condition that strikes patients in the prime of life. The disease robs patients of their ability to carry out everyday tasks (walking, seeing, speaking, interacting, eating, and thinking). There is no currently approved treatment, and with non-specific symptoms at its early stages, PSP is hard to differentiate from Parkinson’s Disease (PD) .

This project explored early diagnosis techniques of PSP, and the objective was to develop a screening protocol for early identification of PSP and differentiation from PD.

 

(Client: Professor, Electrical and Computer Engineering)
Mouse texture cue cube

The project involved the development of an automated texture wheel that can be used to provide differentiated and regulated stimuli to mice while they are running mazes in a virtual reality setting.  The study as a whole is about recreating the electrical network of a human brain, so recording the electrical variances of mice when presented with changes in their dominant sense can help build an electrical mammalian map.  This research strives to understand the electrical signals of the brain for applications to treatment of patients with neurological diseases or injuries.

 

2018 Capstone Projects

(Client: Professor, Electrical and Computer Engineering Dept.)
Systems Genetic Platform of Neurodegenerative Disorders:

Parkinson’s disease is a complex and debilitating neurodegenerative disorder that afflicts over 10 million people worldwide. The Parkinson’s Progression Markers Initiative has compiled, maintained, and distributed an extensive collection of clinical, genetic, and advanced imaging data on Parkinson’s disease. By integrating these complex data, PPMI has offered unparalleled opportunities to investigate the early stages of Parkinson’s, monitor disease progression, and develop novel therapeutics through the identification of progression biomarkers.

Combining complex genetic and imaging data in PPMI, the team sought to explore the use of imaging features and single-nucleotide polymorphisms (SNPs) together as biomarkers for the predictive modeling of Parkinson’s disease. The students proposed, executed, and assessed machine learning approaches for the classification and prediction of Parkinson’s.

 

(Client: Global medical device company)
Intracardiac Electrocardiogram (ICEG) Simulator:

In the current medical device market, there are diverse devices to simulate the physiological signals of the human body, such as surface ECG, SpO2, non-invasive blood pressure, temperature, etc. The ICEG is a type of ECG that measures the cardiac signals inside of the heart through multi-pole catheters that have been weaved into the chambers of the heart. The goal of this project was to design a prototype that could emulate the cardiac signal output taken from 32 channels/signals inside the heart, and create software to measure, analyze, and process these signals. This device would then be used for educational and training purposes and to troubleshoot current or new products.

 

(Client: Start-up medical device company)
Wearable Light Therapy Device for the Treatment of Pain and Nerve Injuries:

The project focused on improving a portable light therapy device developed by the client for consumers and military service members/first responders. The client’s approach was to develop a belt embedded with an array of therapeutic LEDs, which can be worn under clothing and would provide pain relief to the treated area via phototherapy. The team was given the task of solving the heat issues, lack of an automatic shut off and flexibility of the device, while keeping the device lightweight and comfortable to wear.

Additionally, the students wanted to provide patients the added benefit of control over their therapy to create a personalized light therapy device that can be modulated to treat a patient’s unique symptoms. To accomplish this, they incorporated a Bluetooth controller to the micro-controller to allow for mobile monitoring and control over the LED array for personalized therapy.

 

(Client: Start-up medical device company)
Biological Imaging with Synthetic Optical Holography:

The company created an add-on for confocal microscopes using synthetic optical holography (SOH). It is for quantitative phase imaging and allows the user to obtain high-resolution images. This technology results in no loss in speed during image acquisition. It is easy to use and can provide high-quality images without the need to stain.

The goal of this project was to have a working implementation of the SOH technology in the Zeiss LSM 880 confocal microscope located at the Carl R. Woese Institute for Genomic Biology. Also, the team was tasked with testing the SOH technology to determine if there were any problems that needed solving. To do this, the team developed a bank of microscope slides and images that compared phase imaging via SOH with fluorescence imaging.

 

(Client: Global medical device company)
3D Printed Coronaries for a Flow Phantom:

Having standards of known and accurate measurement is useful across multiple scientific disciplines for measuring properties of unknowns and evaluating computational analyses. Phantom vessels that provide realistic representation of human vasculature have been available for decades. While useful for studies that require highly realistic specimens, realistic phantoms generally lack reproducibility and known dimensions, two necessary characteristics of a standard. For this project, the team designed and prototyped phantom blood vessels of simple geometry and phantom coronary artery segments from digital subtraction angiography (DSA) imaging data, with each produced accurately from a 3D computer model stored as a stereolithography (STL) file. These phantom vessels will then be imaged with DSA in a flow loop and used to evaluate measurements performed with an algorithm.

With 3D printing, the students included features present in realistic phantoms (e.g. aneurysms and stenosis), while being able to reproduce phantoms with relatively high accuracy from an STL file.

 

(Client: Regional medical hospital)
Neonatal Jaundice Care for Developing Nations:

The goal of this project was to provide a cost effective, efficient way to treat neonatal bilirubinemia with the development of a fully automated transfusion device. Current methods of treatment for neonatal bilirubinemia are costly, time consuming, and require intensive physician care. In addition, many modern treatment options are unavailable in developing countries because of inhibitive costs, technology, or training. Thus, the team was tasked with designing a device to have the following functionalities and characteristics:

  • Equal extraction and infusion rates
  • Easy to set up
  • Inexpensive and efficient
  • Portable and biocompatible
  • User-friendly interface
  • Blood monitor to ensure patient safety
  • Capped flow rate to avoid excessive pressure on the line
  • Enhanced safety measures to ensure patient care


(Client: Professor, Bioengineering Dept.)

Complete Genome Assembly of Streptococcus sobrinus:

S. sobrinus and S. mutans are the oral pathogens that are responsible for the condition known as caries. S. mutans is identified as being present in all cases of caries but S. sobrinus is without well identified. The focus of this project was to do the complete genome assembly of S. sobrinus strains – 7 and 15. This was done using short read Illumina technology and long read Nanopore technology. The team was also asked to see the genomic similarities between S. sobrinus and S. mutans. The complete genome of S. sobrinus will further help in understanding how genes interact and allow study of metabolic pathways which can be manipulated and redesigned to meet global needs.

 

(Client: Regional Medical Hospital)
Creation of Radiopaque Temporary Embolic:

The goal of this project was to create a radiopaque temporary embolic, or in other words, a device to block blood flow that is visible via x-ray or computed tomography (CT) scan.  Temporary embolics currently used by surgeons tend to blend in with surrounding tissues after insertion, and it is challenging for surgeons to determine their location. Currently, surgeons inject contrast media into the veins of their patients to highlight vasculature in real time under a machine called a fluoroscope. The problem is that this only allows a surgeon to assume the position of an embolic based on the absence of contrast media flow. A radiopaque temporary embolic would allow for the surgeon to quickly and accurately determine the exact location of the embolic throughout and following a procedure.

 

(Client: Professor, Electrical and Computer Engineering Dept.)
Miniaturized Artificial Whisker Scanner and Software:

The project involved the development of a system to simulate mice whisker scanning that also had the ability to read signals related to force in the real mice whisker. Development of such system would allow for better understanding of how the brain works, or more specifically, how the brain perceives the outside sensory world.  This study would also help identify specific neural circuits that are involved in sensory transduction and signal processing. Reverse-engineering of brain circuits can have strong impact on the development of novel biomimetic tactile biosensors, robotic prosthetic arms, haptic virtual reality, and even can influence the design of novel artificial intelligence systems.

 

2017 Capstone Projects

(Client: Regional medical hospital)
Mechanized Bilirubin Scavenging System:

 

A mechanized bilirubin scavenging system for efficient treatment of neonatal jaundice was developed. The unique design uses a bilirubin removal system similar to hemodialysis, where an infant’s blood will be passed through an external scavenging circuit. The overall impact is huge, since exchange transfusion carries a risk of neonatal mortality, especially in sick infants. The adverse effects of an exchange transfusion include neonatal morbidities, such as apnea, anemia, thrombocytopenia, electrolyte and calcium imbalance, risk of necrotizing enterocolitis, hemorrhage, infection, complications related to the use of blood products, and catheter-related complications.

 

(Client:  Regional medical hospital)
Clearing the Clot:

Arterial and venous thrombosis in performing endovascular procedures by interventional radiologists/vascular surgeons/cardiologists is a recurring problem in a clinical setting. The focus of this project was to analyze and distinguish venous and arterial thrombi in a noninvasive and analytical way. The team was also asked to see how do these component mature or change over time, as the thrombus progresses from acute to subacute to chronic. Clinical samples of thrombus/clots from different veins and arteries were collected during re-canalization procedures using different “suction” catheters and mechanical devises. Samples collected were non-invasively analyzed by ultrasound and other techniques.

 

(Client:  Start-up medical device company)
For Your Eyes Only:

This project focused on real-time monitoring of post-surgical and post-traumatic eye injuries using a hand-held device. Lack of current techniques for the early monitoring of bleb leaks and other post-traumatic or post-surgical ocular injury has posed an unmet clinical need for the development of new techniques. Present evaluation techniques use either subjective or non-quantitative approaches. InnSight Technology developed the world’s first biosensor to evaluate the integrity of the anterior surface of the eye by measuring the concentration of ascorbic acid in the tear film at the point-of-care. The team was tasked with developing a tiny micro-fluidic chamber that draws tear fluid from eye to the sensor.

 

(Client:  Integrated providers of diagnostic imaging services)
Project #1 – T1rho Relaxation:
The goal of this project was to simulate the T1rho relaxation effects of an adiabatic RF pulse.  This required the understanding of a rotating frame and its mathematical form, MR RF pulse basics as well as adiabatic design principles. Furthermore, the student studied spin locking and T1rho relaxation using MatLab programming.

Project #2 – iGrasp:
The goal of this project was to get Rapid and Continuous Magnetic Resonance Imaging using compressed sensing, and iGRASP. The student used iGRASP, combining golden-angle radial sampling, parallel imaging and compressed sensing, to reconstruct dynamic MRI image in short time (0.1s). They also focused on using golden-angle radial sampling to get incoherent sampling, which is able to break the limit of Naquist sampling rate that reconstructing by less samples.

 

2016 Capstone Projects

(Client:  Start-up medical device company):
Wirelessly Integrated Ocular Biosensor to Monitor Ascorbic Acid Presence in Tear Film and Aqueous Humor:

Hundreds of eye trauma patients are presented in the emergency department every day. The injuries of the globe can lead to severe eye defects and sometimes vision loss. If the severity of these traumas can be detected early, there can be better recovery of the eye.  After these injuries are treated, postoperative monitoring of eye is very critical to check for any leaks from the anterior globe. If the leak is brisk, the patient has to be taken to the operation room. It is important to detect these leaks as soon as possible so that the vision of the patient is not affected. ­The client has developed a biosensor as a solution to this clinical need.


The principle behind the sensor is that the concentration of ascorbic acid in the aqueous humor is around 20 times the concentration of ascorbic acid in the tear film and when the barrier between them breaks due to any wound or tearing in the corneal epithelium, the concentration of ascorbic acid in the tear film spikes up. This concentration level can be detected by the sensor to get an idea about the severity of the trauma. The biosensor is designed so that when ascorbic acid binds to the enzyme on the sensor, there is a change in the interaction between the polymer and graphene platelets. This changes the electrical properties of the sensors and the change can be measured to get an idea about the injury.  The project objective was to enhance the ability of the biosensor to detect the levels of ascorbic acid.

 

(Client:  Regional medical hospital):
Personalized Absorbable Gastrointestinal Stents for Intestinal Fistulae and Perforations:

Gastrointestinal (GI) tract perforations are relatively frequent surgical emergencies, are potentially life-threatening, and can occur from several different sources, including inflammatory conditions, iatrogenic or traumatic injuries, and obstructive etiologies. Increasing clinical findings corroborate the use of self-expandable metallic GI stents in the setting of gastric or esophageal perforations. Patients admitted to the hospital with intestinal fistulae or perforations typically face months of recovery, unlimited numbers of hospital visits and numerous surgeries that could theoretically benefit from an absorbable stent. Placement of synthetic, non-absorbable stents in the esophagus and colon via endoscopic approaches is limited to these anatomic locations as endoscopic access is required to remove the stents after healing occurs. Commercially produced stents are currently manufactured in a narrow size range of options, further limiting their applicability in other portions of the GI tract.

Initiated by a general surgeon in response to an unmet clinical need, this project objective was to develop novel translatable absorbable polymeric stent, 3D printed for accurate, anatomically personalized placement in the GI tract. In this highly multidisciplinary work, a 3D-printed stent prototype was developed from a novel material using a commercial AirWolf™ device. This functional and effective technology could offer tremendous impact for patients and healthcare providers and significantly reduce patient morbidity and mortality.

 

(Client:  Medical simulation and education center):
Cadaveric Perfusion Pump:
Cadaveric perfusion pumps provide unique opportunities to surgeons and doctors in training. They allow a trainee to practice a surgical procedure under as realistic conditions as is possible before heading into surgery with an actual patient. The pump perfuses a blood mimicking solution through the veins of a cadaver, creating a perfect model for a doctor to practice on.

The client, in collaboration with a regional hospital, has been working to create a perfusion pump that is able to modulate blood pressure and heart rate in order to provide an extra level of realism to the simulations run with the cardiac perfusion pump. The team’s goal was to improve upon the first generation of the client’s artificial perfusion system by making it safer, streamlined and able to accurately generate and measure rapid modulations in fluid pressure. This was done by improving the software to be more robust, improving the setup of the system to be safer and more segregated, finding a pump that is able to generate enough pressure as well as able to switch pressures quickly and integrating sensors into the code that are used to ensure the proper operation of the system as well as act as a safety check.